Learn AI in 5 Minutes a Day

AI Tool Report is one of the fastest-growing and most respected newsletters in the world, with over 550,000 readers from companies like OpenAI, Nvidia, Meta, Microsoft, and more.

Our research team spends hundreds of hours a week summarizing the latest news, and finding you the best opportunities to save time and earn more using AI.

Sign up with 1-Click

Today:

• HeyGen Simplifies AI Video Creation

• OpenAI Erases Crucial Evidence Accidentally

• OpenAI Challenges Google with Browser

• Snowflake Integrates Claude, Beats Databricks

• Black Forest Labs Launches FLUX.1

• Brave Search Introduces AI Chat

AI Video is getting UNREAL... plus other AI News.

HeyGen's iOS app simplifies creating AI avatar videos, making advanced technology accessible. It allows users to generate videos and lifelike avatars with text or audio inputs. 

With support for 175 languages, it’s a game-changer for content creators. Users can train the app using personal images or videos to produce personalized avatars that mirror their voice and expressions. 

WATCH THE VIDEO ON YOUTUBE

OpenAI Erases Crucial Evidence Accidentally

OpenAI engineers mistakenly erased crucial evidence in a lawsuit filed by The New York Times and other publishers over AI training data usage. The legal teams spent over 150 hours identifying their articles in OpenAI’s data, but the erasure made the findings incomplete and unreliable. The lawsuit alleges OpenAI and Microsoft used millions of articles to train their AI, competing directly with publishers. 

The Times seeks billions in damages, claiming unfair practices. OpenAI calls the incident a "glitch" and denies wrongdoing. Meanwhile, OpenAI partners with other media companies, suggesting publishers are divided between legal battles and collaboration.

OpenAI Challenges Google with Browser

OpenAI is exploring ventures to challenge Google, including developing a browser integrated with its ChatGPT chatbot. The company has also discussed powering AI-driven search features for industries like travel, food, real estate, and retail, collaborating with brands like Condé Nast, Redfin, and Priceline. 

Additionally, OpenAI has explored partnerships to enhance AI functionalities on devices by companies like Samsung, echoing its recent collaboration with Apple. These moves signal OpenAI's ambitions to expand its influence in search and AI applications across diverse platforms, directly competing with Google in key markets.

Snowflake Integrates Claude, Beats Databricks

Snowflake has partnered with Anthropic to integrate its Claude 3.5 AI models directly into Snowflake’s data cloud platform on AWS. This collaboration enables Snowflake’s 10,000+ customers to build advanced AI applications, such as conversational agents and data analytics tools, within a secure, governed environment. 

Unlike competitor Databricks, which offers Claude models via API, Snowflake provides direct integration for seamless deployment. The Claude models will also enhance Snowflake’s internal workflows and customer-facing tools like Snowflake Intelligence. By simplifying AI application development and reducing configuration time, Snowflake aims to attract more businesses seeking efficient, enterprise-grade AI solutions.

Black Forest Labs Launches FLUX.1

Black Forest Labs has launched FLUX.1 Tools, enhancing their FLUX.1 text-to-image model with four new features:

• FLUX.1 Fill: Advanced inpainting and outpainting for seamless image edits and extensions.

• FLUX.1 Depth and Canny: Structural guidance using depth maps or canny edges for precise transformations.

• FLUX.1 Redux: Image variation and restyling using input images and prompts.

These tools are available as open-access models under the Flux Dev License and through the BFL API for professionals. Benchmarks reveal state-of-the-art performance across applications, supporting tasks like retexturing, editing, and image recreation. Partners include fal.ai, Replicate, Freepik, and more.

Brave Search Introduces AI Chat

Brave has introduced an AI chat feature to its Brave Search platform, enabling users to ask follow-up questions based on initial search queries. Building on its “Answer with AI” summaries, this new feature integrates chat functionality directly into search results, offering a unique experience not available on Google. 

Users can ask related questions seamlessly, accessing AI-generated answers supported by source citations. The feature leverages a mix of open and internal language models while prioritizing user privacy by not storing queries. Brave Search processes over 36 million queries daily, with AI tools providing more than 11 million answers, enhancing its competitive edge.

🧠RESEARCH

SageAttention2 Technical Report: Accurate 4 Bit Attention for Plug-and-play Inference Acceleration

The SageAttention2 report introduces a faster, more efficient 4-bit attention mechanism for AI models, achieving 3-5x speedups over competitors like FlashAttention2. It ensures precision with innovative quantization techniques and smoothing methods, offering near-zero accuracy loss across tasks like language processing, image, and video generation. Open-source code is provided.

VBench++: Comprehensive and Versatile Benchmark Suite for Video Generative Models

VBench++ offers a comprehensive benchmark for evaluating video generative models, addressing gaps in human-aligned metrics and model insights. It dissects video quality into 16 detailed dimensions, evaluates text-to-video and image-to-video tasks, and provides open-source tools to advance video generation research. Human preferences validate its relevance and trustworthiness.

VideoAutoArena: An Automated Arena for Evaluating Large Multimodal Models in Video Analysis through User Simulation

VideoAutoArena introduces an automated benchmarking framework for evaluating multimodal models in video analysis. It simulates user interactions, generates adaptive questions, and uses a scalable ELO-based system to assess model performance. Aligned with human judgments, it offers cost-effective insights into model strengths, evolving challenges, and areas for improvement.

SAMURAI: Adapting Segment Anything Model for Zero-Shot Visual Tracking with Motion-Aware Memory

SAMURAI enhances SAM 2 for zero-shot visual tracking by integrating motion-aware memory selection, improving object tracking in dynamic scenes without retraining. It achieves real-time performance, surpassing existing trackers in success and precision across benchmarks, demonstrating robust, adaptable tracking in complex environments.

ViBe: A Text-to-Video Benchmark for Evaluating Hallucination in Large Multimodal Models

ViBe introduces a benchmark to evaluate hallucination in text-to-video (T2V) models, categorizing issues like vanishing subjects and temporal errors. With 3,782 annotated videos, it advances detection and mitigation techniques, promoting more reliable T2V outputs.

🛠️TOP TOOLS

Dashworks - Create Custom AI Search Assistants

Reword - Write people-first articles with an AI trained by you.

Unriddle - Quickly find info in research papers, simplify complex topics, write with AI and keep everything organized.

Flick - AI-powered social marketing platform that help with scheduling, hashtags, caption writing, and analytics.

Theneo - Generate Stripe-like docs in seconds

📲SOCIAL MEDIA

🗞️MORE NEWS

• Huawei plans to mass-produce its advanced AI chip, Ascend 910C, by early 2025 despite U.S. trade restrictions. However, production challenges and low chip yields hinder its competitiveness with global rivals.

• Pony AI seeks a $4.55 billion valuation in its U.S. IPO, increasing share offerings amid growing interest in Chinese robotaxi firms. It marks a significant move despite U.S.-China economic tensions.

• Nvidia's quarterly revenue surged 94% to $35.1 billion, driven by AI chip demand. Despite overheating issues with new Blackwell chips, sales exceeded forecasts, highlighting Nvidia's critical role in AI's growth.

• Elon Musk encouraged X users to upload medical scans, such as MRIs and CT scans, to train its AI chatbot, Grok, in interpreting medical data, sparking debate over data privacy and AI development.

• Salesforce's new Agentforce Testing Center allows enterprises to evaluate AI agents in controlled environments, testing workflows, monitoring performance, and ensuring alignment with business needs.

• YouTube Shorts’ Dream Screen now generates AI video backgrounds using Google DeepMind’s Veo model, allowing creators to customize cinematic settings. This feature surpasses TikTok by offering video, not just image, backgrounds.